ContentID Technology

Anvato combines its patent pending computer machine vision (Perceptual Signature™) with scalable vision search (Perceptual Search™) to search and identify videos based on their content. Unlike DRM (Digital Rights Management) techniques, our approach does not rely on encryption or watermarking. If a human being can watch the video, we can index and search for it.

Perceptual Signature™

Anvato processes videos (images and sounds) to extract a searchable, sortable sequence of numbers called Perceptual Signature. This signature simulates the human perception of the video. It not only accurately and uniquely identifies the video and its content, but also it is very small and easy to search.

Perceptual Signature is not a mathematical artifact or a checksum; it contains unique information about the context of the video, such as scenes, objects and how these objects move.

Perceptual Search™

Perceptual Search rapidly and accurately identifies how visually similar two videos are. Perceptual Search does not require the videos to be identical; its identification is based on visual similarity of content. Perceptual Search offers the following unique benefits.

• Scale and Speed. Searches about one million minutes of video under five seconds using a six computer grid. Scale and speed increase as the grid size increases.
• Robustness. Captures the contextual information from videos, including the scenes, the objects, and their movements. It is inherently robust against:
  • Missing or replaced sound
  • Transcoding (e.g., format conversion FLV/ MPEG)
  • Image degradation or compression
  • Geometric distortion (e.g., aspect ratio change from 16:9 to 4:3)
  • Video editing attacks such as partial blocking and mashups